Somewhat random BSOD &/or Graphics driver crash. (Solution included)

[SolarNova]

Greetings.

 

So for almost a year I had been putting up with what seemed like random and somewhat infrequent crashing (until i tried to play certain games, then it was very frequent). This usually occurred during gaming , but every now and then happened outside of gaming.

The crashes came in 2 forms. Graphics driver crashes that could sometimes be recovered from, or an outright BSOD.

 

Once i finally decided to deal with the problem these are the initial thoughts and steps i took to narrow down the problem.

It should be noted this, from what i could recall at the time this seemed to have started after redecorating and reorganizing the room I'm using

 

Initial thoughts and steps:

 

Faulty extension lead (multiplug) - swapped it out with a spare.

Bad wall socket - switched to a different one

Loose wire inside case - Checked wires, gave them a wiggle etc.

Faulty case power button - possibly knocked during decorating - disconnected from motherboard.

Failing PSU - couldnt check, no spare

Corrupted OS - couldnt immediately check due to lack of storage space

Possible failing GPU - worst case scenario (price)

 

By this point it was becoming more of an issue and i really needed it fixed, so for Xmas I told my family that for a present they could between them buy me a 1TB SSD so i could install a fresh version of Windows and hopefully that would fix my problem. At the very least it would give me some much needed storage space.

 

So Xmas came, i got my drive, i installed a version of W10 and....

 

It seemed to have fixed the issue, right up until i got 5 minutes in to Path of Exile ( a game that was crashing almost instantly)

However it was not crashing in MechWarrior Online.

After a series of benchmark runs without fail, i figured POE was just running a unstable patch.

Then a benchmark i had running whilst i was eating dinner failed.

 

At this point i was seriously considering a failing GPU as the cause.

 

However at this point i had also considered one other thing, one that i honestly didnt think was likely.

A Crap GPU riser cable. - This would be a right PITA to test as it would require draining my water loop which use exceedingly tight fitting soft tubing and compression fittings and no drain port (my own fault)

This was a possibility because I had also in the last year, rebuilt my loop so that i could vertically mount my GPU, and thus bought a vertical mount with a riser cable.

But i figured, this is a PCI-E 3.0 system, an old Sandybridge platform, with a 1080ti, its not like these 4.0 risers that are currently failing, could it really be the riser ?

 

So, I decided I would test both the GPU and the riser theory at the same time, I would remove my GPU, put the old aircooler back on it, re plum by loop for CPU only, put the GPU back  directly into the PCI-E slot, and test. If it failed it wasnt the riser and i could remove the GPU and put it into another computer in the house and try it on that system to ensure its the GPU and not something else in my system.

 

Then Mr Sod reared his head (Sods law), My 1080ti i bought is 2nd hand (used), it came with the water block i needed preinstalled. Turns out the OG air cooler which was also included has terrible mounting design (the white and orange Gigabyte OC version), near impossible to attached once removed from the PCB, The built in 'standoffs' of the block need to pass through the PCB, but when originally fitted the standoffs expand like a drywall rawplug, thus making it next to impossible to reattached once removed.

So with an incredibly bad 'mount' i tested the GPU, it ran hot, real hot.. but i could run it long enough to test.

 

Go figure ! it worked.. no crashing !

 

All this hassle, and it turned out to be the damned cheap vertical GPU mount ..or rather, the riser cable.

 

I have now reattached the waterblock, drained my loop again and re-plumed it. Not only is it no longer crashing but thanks to a fresh repaste, its running cooler (not that it needed to)

 

 

TLDR:

 

Dont use cheap PCI-E riser cables !

 

UPDATE:

It wasnt the riser cable ! Go figure, the issue returned, just to a lesser extent.

Turned out it was my PSU, replaced and problem solved permanently this time.